Study population Potential eligible subjects will be identified from the community via several sources, including but not limited to existing cohorts of subjects who previously participated in other study or health education talk and agreed to be contacted for clinical research recruitment; such as users of smartphone-based hypertension management tool (HealthCap), and members from CUHK Jockey Club Community Primary Care Programme and S.H. Ho Centre for Digestive Health. The subjects can choose to be screened for eligibility and obtain informed consent on WhatsApp Messenger/telephone or face-to-face study visits. Two $50 cash coupons incentives will be mailed to subjects who participate (upon successful recruitment) and complete this study (upon completion of the 6-month telephone follow-up) to speed up recruitment.
The inclusion criteria for this study are as follows: 1) 50-75 years old; 2) asymptomatic and eligible to enrol in a government-subsidized colorectal cancer screening program (26); subjects themselves or their family members living in the same household with access to WhatsApp Messenger; 4) able to read Chinese; and 5) willingness to be followed up by telephone. Those without access to WhatsApp Messenger or unable to provide informed consent will be excluded. Subjects who have incomplete conversation with the chatbot will also be excluded.
Sample size estimation Based on the local baseline CRC screening uptake rate of 10% and the results from a recent meta-analysis showing that the risk ratios for CRC screening uptake of subjects receiving patient reminder and outreach were 1.2 and 2.2 respectively when compared to subjects in the control group (23), we assume that subjects in the standard text reminder group will have a 12% screening uptake rate after receiving a text message reminder; and subjects in the chatbot outreach group will have a 22% screening uptake rate. A minimum sample size of 436 subjects (218 per group) was required to achieve a statistical significance of p \< 0.05, and a power of 80%. Assuming 10% of subjects will drop out, lose follow-up, or have incomplete conversation with the chatbot. The final sample size is rounded up to 500.
All eligible subjects will be randomized and allocated to receive either intervention (chatbot) or standard of care (text reminder) in a 1:1 allocation ratio. A fully automated chatbot will be developed and maintained in collaboration with IT experts on the WhatsApp Messenger platform for its wide popularity in local older adults and user-friendly interfaces for extensions. Depending on the randomization group, it delivers either a standard text reminder of CRC screening or health education in text and video formats with a series of questions on personalized risk assessment of CRC and information of primary care physicians under a government-subsidized CRC screening program according to the subject's preferred district. The same professional team will maintain the health status of the chatbot during the intervention period. A monthly review of chatbot performance will be conducted. A feedback loop is also incorporated into the chatbot, and users are invited to leave comments. Accordingly, improvements will be made, and the questions raised by the subjects will be collected and used to construct a data bank for the future development of Artificial intelligence (AI)-enabled chatbots.
Outcome measures The primary outcome is the CRC screening uptake rate, defined as participation in and completion of the government-subsidized CRC screening program by returning valid FIT kits at three months after intervention.
The secondary outcomes include 1) CRC screening uptake rate at six months after recruitment, 2) baseline intention of CRC screening before intervention, measured by 4 point Likert scale (1 = strongly disagree; 2 = somewhat disagree; 3 = somewhat agree and 4 = strongly agree), 3) change in screening intent immediately after each intervention, evaluated by calculating point difference between pre- and post-intervention, 4) time interval to participate (Date of FIT kits collection minus date of study intervention) and complete screening after recruitment (Date of return of valid FIT kit minus date of study intervention), and 5) reason for not participating in screening (common local barriers to CRC screening, including, financial difficulty, limited services accessibility, screening-induced bodily discomfort, physical harm, embarrassment, apprehension, time constraints and others)
Telephone follow-up All subjects will be interviewed by phone three and six months after recruitment and asked if they have participated in the government-subsidized CRC screening program and CRC screening completion rate. The date of participation, completion of CRC screening, and FIT results will also be assessed. The reasons for not participating in the CRC screening program will be documented. We will also ask the subjects who have completed CRC screening if our chatbot or text reminder motivated them to uptake or there is exposure to any possible confounders, e.g., community initiative, related symptom onset, healthcare worker advice and family member or friend initiative, etc. A cash coupon of $50 Hong Kong Dollar will be mailed to the subject upon completion of telephone follow-up.
Focus group discussions 12 focus group discussions (6-8 participants per group) will be conducted after the telephone follow-up. The purpose of the focus group discussions is to identify the process factors to understand how the study intervention influence screening uptake behaviors and explore the complexity of FIT screening and reveal, based on the participants' accounts, the barriers, and facilitators for them to uptake FIT screening. Six groups from each RCT arm will be convened. For each discussion session, an experienced moderator from our research team will introduce our study aims, lead the discussion in a semi-structured format based on the discussion protocol, and ask clarifications when necessary. Another assistant will take notes and address contingent issues. Each focus group will be audiotaped and last for 1.5-2.5 hours. All participants will sign a written consent form before the focus group session and will be given some incentives at the end of the session to compensate for their time.
Quantitative analysis Microsoft ACCESS, version 2016 (Redmond, Washington, USA), Statistical Analysis Software, SAS (Cary, NC, USA) and the Statistical Package for Social Sciences (SPSS), version 28.0 (Chicago, IL, USA), will be used for all data entry and analysis. Continuous variables were described in terms of their means and standard deviations (SD) and analyzed using Student's t-test. Pearson's chi-squared test or Fisher's exact test will be used to assess categorical variables, when appropriate. Statistical significance will be set at P \< 0.05. Intention-to-treat (ITT) analysis included all recruited patients. Per-protocol (PP) analysis included only subjects who completed telephone follow ups and chatbot conversations.
Qualitative analysis To explore and identify various perspectives on nonadherence, thematic analysis will be applied to analyze data obtained from the focus groups. The principal investigator and research assistant will separately create codes for approximately 20% of the transcripts, and they will jointly agree upon a common code book. As they review the transcripts separately, they will repeatedly compare their coding schemas to identify new themes and codes, ultimately developing a consensus-based interpretation of the data. This iterative process ensures consistency in the data analysis and generates a comprehensive understanding of the ideas presented throughout the focus group discussions. Dedoose (Manhattan Beach, California, USA) will be used for qualitative data analysis.
